DRAFT

2008 ERCOT LOADS ACTING AS A RESOURCE

CAPABILITY STUDY

DECEMBER 2008 PREPARED FOR THE ERCOT RELIABILITY AND OPERATIONS SUBCOMMITEE

BY THE ERCOT DYNAMICS WORKING GROUP DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

TABLE OF CONTENTS

STUDY GROUP...... 1 DISCLAIMER...... 2 EXECUTIVE SUMMARY...... 3 INTRODUCTION...... 3 STUDY OBJECTIVE...... 3 STUDY RESULTS...... 3 INTRODUCTION...... 5 STUDY OBJECTIVE...... 5 BACKGROUND – LaaRs, GENERATION SPINNING RESERVES AND GENERATION GOVERNING...... 5 STUDY METHODOLOGY...... 6 Software...... 6 Models and Data...... 6 Network Model Data...... 6 Dynamics Model Data...... 7 Governor Models...... 7 LaaR Modeling...... 9 Low Set Underfrequency Relay Modeling...... 10 Load Modeling...... 10 Wind Generation Modeling...... 11 Study Approach...... 12 Simulation Plan...... 12 Other Simulation Details...... 16 Study Criteria...... 16 STUDY RESULTS...... 17 Summer Peak Conditions...... 17 Spring Off-Peak Conditions...... 17 Series 3...... 17 Series 4...... 18 Series 5...... 18 Series 6...... 18 Series 7-10...... 18 Series 11-14...... 19 Series 15-18...... 19 Series 19-22...... 20 Series 23-30...... 20 Series 31-35...... 20 Series 36...... 21 Series 37...... 22 CONCLUSIONS...... 23

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

i DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

STUDY GROUP 2008 Dynamics Working Group DWG MEMBER COMPANY Anthony Hudson, Chair Texas-New Mexico Power Company Jose Conto ERCOT System Planning John Schmall ERCOT System Planning David Milner, Vice Chair City Public Service Vance Beauregard American Electric Power Roy Boyer Oncor Reza Ebrahimian Austin Energy Shun-Hsien Huang ERCOT Operations Support David Mercado Centerpoint Energy Tom Bao Lower Colorado River Authority John Moore South Texas Electric Cooperative

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

1 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

DISCLAIMER The Electric Reliability Council of Texas (ERCOT) Dynamics Working Group prepared this document. Conclusions reached in this report are a “snapshot in time” that can change with the addition, or elimination, of plans for new generation, transmission facilities, equipment, or loads. ERCOT AND ITS CONTRIBUTING MEMBER COMPANIES DISCLAIM ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING ANY WARRANTY OF MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE WHATSOEVER WITH RESPECT TO THE INFORMATION BEING PROVIDED IN THIS REPORT. The use of this information in any manner constitutes an agreement to hold harmless and indemnify ERCOT, its Member Companies, employees and/or representatives from all claims of any damages. In no event shall ERCOT, its Member Companies, employees and/or representatives be liable for actual, indirect, special, or consequential damages in connection with the use of this data. Users are advised to verify the accuracy of this information with the original source of the data. The Electric Reliability Council of Texas (ERCOT) manages the flow of electric power to 21 million Texas customers – representing 85 percent of the state’s electric load and 75 percent of the Texas land area. As the independent system operator for the region, ERCOT schedules power on an electric grid that connects 38,000 miles of transmission lines and more than 550 generation units. ERCOT also manages financial settlement for the competitive wholesale bulk-power market and administers customer switching for 6 million Texans in competitive choice areas. ERCOT is a membership-based 501(c)(4) nonprofit corporation, governed by a board of directors and subject to oversight by the Public Utility Commission of Texas and the Texas Legislature. ERCOT's members include consumers, cooperatives, independent generators, independent power marketers, retail electric providers, investor-owned electric utilities (transmission and distribution providers), and municipal-owned electric utilities.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

2 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

EXECUTIVE SUMMARY INTRODUCTION In October 2007, the ERCOT Technical Advisory Committee (TAC) approved a proposal to change the Responsive Reserve Service (RRS) obligation from the current 2300 MW to 2800 MW. A current ERCOT operating guide rule allows for 50% of the ERCOT RRS obligation to be composed of Loads Acting as a Resource (LaaR) tripped at 59.7 Hz, with the remainder being spinning reserves. The current rule, however, is based upon the previous RRS obligation of 2300 MW. This translates to a maximum allowable 1150 MW of LaaRs tripped at 59.7 Hz. At this maximum level of LaaRs, there remains a substantial spinning reserve pool of 1650 MW under the new 2800 MW RRS obligation. ROS assigned this study to the Dynamics Working Group (DWG) to answer questions originally raised by the ERCOT Long Term Solutions Task Force (LTSTF). The LTSTF asked whether reliability concerns would be raised by increasing LaaRs, tripped at 59.7 Hz, based upon the new RRS obligation of 2800MW. Furthermore, the LTSTF asked how much higher a LaaR maximum limit is possible if frequency tiered deployment were considered. STUDY OBJECTIVE There are two objectives of this study. The first objective is to determine the LaaRs percentage of 2800 MW RRS obligation, tripped at 59.7 Hz, where reliability concerns are raised. The second objective is to determine the incremental LaaRs percentage of 2800 MW RRS obligation, tripped at 59.8 Hz, where reliability concerns are raised. STUDY RESULTS Because system inertia is at the lowest levels during light loading, conditions are optimal for system frequency overshoot. Consequently, the critical operating scenarios for reliable deployment of LaaRs are during periods of light loading. The most significant variables to consider in determining the maximum amount of LaaRs that can be deployed via underfrequency relaying are the frequency trip settings of the LaaR relays. The worst case scenarios for frequency overshoot occur when all of the LaaRs trip at the same frequency set- point. Intermediately, frequency overshoot can be depressed to varying degrees when the frequency set-points of the LaaR relays are spread out over a range of frequencies. Other variables having a more limited relationship with the maximum amount of LaaRs deployable by underfrequency relaying are the geographical locations of the LaaRs. Table 1 summarizes the various limits on the amount of LaaRs deployable at the 59.7 Hz tier and the conditions associated with each limit.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

3 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Table 1 – Summary of LaaRs Limits at the 59.7 Hz Tier RELAY TRIP SETTING LaaRs LIMIT (% of 2800 LOCATION OF LaaRs LIMIT BASIS (Hz)† MW) Current trip settings: 9.1% @59.8 Hz 0.9% @59.72 Hz 1.4% between 59.71 and Current locations 59.72 Hz 65 10.7% @59.71 Hz 25.4% between 59.70 and 59.71 Hz 52.5% @59.70 Hz Modified version of the current trip settings: 0.9% @59.72 Hz 1.4% between 59.71 and Overshoot greater than Current locations 59.72 Hz 60 or equal to 60.4 Hz. 10.7% @59.71 Hz 25.4% between 59.70 and 59.71 Hz 61.6% @59.70 Hz Current locations 59.72 55 Uniformly distributed 59.74+ 55 throughout ERCOT Lumped in NTX CSC zone 59.72 55 Lumped in STX CSC zone 59.74 50 Lumped in WTX CSC zone 59.72 50 Lumped in HOU CSC zone 59.74+ 50 †The relay trip setting data in row 1 of this table is representative of the current LaaRs in ERCOT. The relay trip setting numbers in rows 3 through 8 are maximum LaaR trip frequencies assuming all of the LaaRs specified in the “LaaRs LIMIT” column are tripped at the same frequency. Results from simulations of LaaRs deployed at the 59.8 Hz tier indicate such LaaRs cannot be reliably deployed with the 59.7 Hz tier LaaRs operated at the critical levels identified in table 6. The maximum amount of LaaRs and corresponding maximum relay trip settings that can be reliably deployed at the 59.8 Hz tier is a function of where limits are established for the amount of LaaRs and corresponding relay trip settings at the 59.7 Hz tier. Therefore, the DWG will need guidance from ROS on where the 59.7 Hz tier limits will be set in order to define LaaR limits for the 59.8 Hz tier. Section 6.10.3.2 of the ERCOT Protocols allow a LaaR to be deployed at up to 150% of the amount requested by ERCOT at the time of testing the LaaR. This potential variance between contracted and actual LaaR amount should be considered when establishing LaaR limits.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

4 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

INTRODUCTION In October 2007, the ERCOT Technical Advisory Committee (TAC) approved a proposal to change the Responsive Reserve Service (RRS) obligation from the current 2300 MW to 2800 MW. A current ERCOT operating guide rule allows for 50% of the ERCOT RRS obligation to be composed of Loads Acting as a Resource (LaaR) tripped at 59.7 Hz, with the remainder being spinning reserves. The current rule, however, is based upon the previous RRS obligation of 2300 MW. This translates to a maximum allowable 1150 MW of LaaRs tripped at 59.7 Hz. At this maximum level of LaaRs, there remains a substantial spinning reserve pool of 1650 MW under the new 2800 MW RRS obligation. ROS assigned this study to the Dynamics Working Group (DWG) to answer questions originally raised by the ERCOT Long Term Solutions Task Force (LTSTF). The LTSTF asked whether reliability concerns would be raised by increasing LaaRs, tripped at 59.7 Hz, based upon the new RRS obligation of 2800MW. Furthermore, the LTSTF asked how much higher a LaaR maximum limit is possible if frequency tiered deployment were considered. STUDY OBJECTIVE There are two objectives of this study. The first objective is to determine the LaaRs percentage of 2800 MW RRS obligation, tripped at 59.7 Hz, where reliability concerns are raised. The second objective is to determine the incremental LaaRs percentage of 2800 MW RRS obligation, tripped at 59.8 Hz, where reliability concerns are raised. BACKGROUND – LaaRs, GENERATION SPINNING RESERVES AND GENERATION GOVERNING Whenever generation is not in balance with the total demand, the electrical frequency of the entire interconnect will deviate from the nominal 60 Hz frequency at which the system was designed to operate. So the total generating capacity in a power system must be sufficient to supply the expected peak load demand plus a margin, or operating reserve. On a daily basis a system must carry enough operating reserves to regulate and to allow for unanticipated events, including forced outages and load forecast errors. Operating reserves are comprised of spinning reserves, non-spinning reserves, LaaRs, and DC tie-line response. Spinning reserves are generation operating at less than peak output, which are synchronized and immediately respond to frequency changes. LaaRs are loads that are tripped, either by frequency relaying or by operator action, to mitigate underfrequency events within the ERCOT system. Responsive Reserves are a subset of operating reserves which ERCOT maintains to restore system frequency within the first few minutes of an event. Small load variations take place all the time, so frequency continuously deviates from 60 Hz. These smaller variations in frequency are covered by regulating reserve, which is made up of the portion of spinning reserve responsive to automatic generation control. These normal frequency deviations are quite small compared to those that occur following large disturbances. In addition to the considerable deviation in frequency from 60 Hz large disturbances can impose, an interconnected system will have natural system oscillations in frequency following a system disturbance. This oscillatory condition is normally damped in a large system but damped to a lesser extent, or even unstable, in a lightly loaded system with lower inertia. The focus of this study is the effect on the system response during the first 15 seconds following a large disturbance from varying two components of Responsive Reserves; spinning reserves and This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

5 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

LaaRs. The particular group of LaaRs that are investigated in this study are those that are tripped by relaying within 30 cycles for system frequencies below a predetermined frequency set point, currently no less than 59.7 Hz in the ERCOT system A key mechanism to spinning reserves is generator governing. Governing is the automatic response of the governor to a usually large change in frequency during the first 15 seconds following an event. Governing is the process where a generating unit changes its power output in response to a change in frequency. If the frequency drops below 60 Hz, governing will increase generation power output to arrest the frequency decline. Alternatively, if the frequency increases above 60 Hz, governing will decrease generation power output to arrest the frequency rise. For governing response to be effective, the following three elements must exist: 1. The unit must have a governing margin. If the unit is operating at full load, it cannot increase its output in response to a loss of generation. Similarly, a unit operating at 90% of full load cannot respond with 20% of its capacity to a loss of generation. 2. The unit’s controls must permit governing. “turbine follow” and “sliding pressure control” for conventional steam plants, and operating combustion turbines on temperature control can effectively block governing action. 3. The unit must have a governor or speed input to the plant controls that is not blocked by intentional dead-bands or limiters. The rate and magnitude of governor response to a speed change can be tuned for the characteristics of the generator that the governor controls and the power system to which it is connected. STUDY METHODOLOGY Software The Siemens PTI Power System Simulator for Engineering (PSSE) software was used for all simulations in this study. Models and Data Network Model Data To capture the approximate ERCOT operating extremes, in terms of magnitude of load and generation, two network model cases were used as a starting point in the analysis; a summer peak case and a spring off-peak case. The following summer peak case was used: 08SUM1 -2008 SUMMER ON-PEAK BASE CASE - ERCOT ROS SSWG UPDATED 08/31/2007 - ERCOT PSSE VER 30.3 CSC CONS. DISPATCH The following spring off-peak case was used: 08SPG2 -2008 SPRING OFF-PEAK BASE CASE - ERCOT ROS SSWG UPDATED 11/28/2007 - ERCOT PSSE VER 30.3 CSC CONS. DISPATCH In each of the above two network cases, online non base load generating units and combined cycle trains from across the entire ERCOT system were selected for simulation of the spinning reserve portion of responsive reserves. These units are

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

6 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

herein referred to as “participating units”. In order to meet the objectives of this study, it was necessary to simulate a range of LaaRs and corresponding generation spinning reserve combinations. This was facilitated by altering generation dispatch of various participating units, creating numerous variations of each of the two starting network cases. These network case variations are summarized in Table 2. Details on LaaRs modeling is discussed in the “Dynamics Model Data” and “Study Approach” sections below. Table 2 – Network Case Variations Created For The Study AMOUNT OF SPINNING RESERVE MIX OF SIMPLE CYCLE AND MODELED IN THE NETWORK COMBINED CYCLE UNITS NETWORK CASE1 CASE SIMPLE CYCLE COMBINED % OF 2800 MW MW (MW) CYCLE (MW) RRS SUM SPG SUM SPG 40% LaaRs 60 1680 1110 1039 570 641 45% LaaRs 55 1540 970 934 570 606 50% LaaRs 50 1400 913 812 487 588 55% LaaRs 45 1260 798 721 462 539 60% LaaRs 40 1120 704 673 416 447 65% LaaRs 35 980 589 570 391 410 70% LaaRs 30 840 520 494 320 346 75% LaaRs 25 700 397 393 303 307 Dynamics Model Data Governor Models In the vast majority of the simulations performed for this study, governor models are used for the participating units while governor models are not used for the non- participating units. This effectively simulates the limitation of governor response to the participating units. While this is a conservative modeling approach for summer peak conditions, it was not intuitive whether or not this approach would be conservative for spring off-peak conditions. Therefore, a number of simulations were performed using the spring off-peak cases with governor models used for all on-line units. Details on the set-up for these simulations are provided in the sections that follow. From previous comparisons of recorded frequency data following large disturbances within the ERCOT system to corresponding simulation results using “as-is” governor model data from the ERCOT Dynamics Model Database, it has been consistently demonstrated that the recorded frequency response does not match well with the simulated frequency response. Specifically, the ERCOT system has always been, in actuality, less responsive than the simulated ERCOT system. There have been a number of plausible explanations for these differences in response. Improvements in combined cycle governor modeling (combustion turbine and steam turbine governor models) over the last couple of years have significantly narrowed the gap between actual and simulated frequency response. However, more work is needed in this area to determine the remaining governor model inaccuracies and how to address them.

1 In compliance with the ERCOT Operating Guide, all participating units are dispatched at no less than 80% of capability in all of the network cases used for this study. For participating combined cycle trains, all trains are dispatched at no less than 80% of capability. This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

7 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Given the discrepancy in recorded and simulated frequency response discussed above, the DWG realized some level of governor tuning would be necessary in order to produce sound results for this study. A method that was considered involved tuning governors on a unit basis utilizing unit specific power output data in response to actual events. In this way, unit specific characteristics, particularly whether or not the unit utilizes load control (runback), could be appropriately modeled. Because unit specific event data is not currently made available with corresponding system event data, this method proved to be infeasible. Alternatively, the DWG used a familiar governor tuning method used in previous DWG studies. As opposed to approximating the response of generating units on an individual basis, this method approximates the average response of generating units on an aggregate basis. The general steps for this method involved: 1. Choose a single governor model (or generic governor model) to replace the governor model for each simple cycle participating unit. Governor models for the participating units utilizing the new combined cycle governor models were not replaced with the generic governor model. 2. Adjust the droop and certain time constants of both the generic governor model and the combined cycle gas turbine governor model such that the simulated system frequency response for two or more known events approximates the recorded frequency response for those events. Because of its relative simplicity and common use within the ERCOT system, the IEESGO governor model from the Siemens PTI model library was selected for use as the generic governor model. Although the tuning process turned out to be rather complex, event data from only two events was sufficient for obtaining tuned generic governor models. The 8/19/04 loss of the Forney plant and the 2/3/08 loss of Martin Lake 1 events were used. Both recorded and simulated frequencies for the Forney and Martin Lake events are shown in figures 1 and 2, respectively. The IEESGO and combustion turbine governor model parameters used in this study are shown in figure 3. Figure 1 – 8/19/04 Forney Plant Trip, Recorded and Simulated Frequencies Using Tuned Governor Models

Frequency vs. Time

60.05

60 ) z H

( Recorded

q 59.95

e Simulated r F 59.9

59.85 -5 0 5 10 15 20 25 30 35 Time (sec)

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

8 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Figure 2 – 2/3/08 Loss of Martin Lake 1, Recorded and Simulated Frequencies Using Tuned Governor Models

60.05

60

59.95 LaaR Trip Simulated 59.9 Recorded

59.85

59.8

59.75 -5 0 5 10 15 20 25 30 35

Figure 3 – Final Generic Governor Model Parameters

LaaR Modeling For dynamic simulations, the simulation of automatic LaaR tripping is accomplished through the use of an underfrequency load shedding model. The desired frequency trip setting, trip delay, and breaker operate time is entered into the model for any particular load or group of loads. In this study, there were a myriad of LaaR model configurations utilized to establish the set-up for: 1. Various LaaR and spinning reserve combinations analyzed, 2. Various LaaR locations analyzed, 3. Various LaaR frequency trip points analyzed, and 4. Various LaaR trip delays analyzed. The specific LaaR configurations used in this study are outlined in the “Study Approach” section below.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

9 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Low Set Underfrequency Relay Modeling Section 2.9 of the ERCOT Operating Guide requires at least 25% of the ERCOT system load that is not equipped with high set underfrequency relaying (LaaRs) to be equipped with automatic underfrequency relaying. The firm load shed amounts and corresponding frequency thresholds are as follows: o 5% of the ERCOT system firm load at no less than 59.3 Hz. o An additional 10% of the ERCOT system firm load, for a total of 15%, at no less than 58.9 Hz. o An additional 10% of the ERCOT system firm load, for a total of 25%, at no less than 58.5 Hz. Considering the capacity of the largest single ERCOT generating plant (2500 MW), the amount of RRS being simulated (2800 MW), the relatively smaller amounts of generation tripping being simulated in the spring off-peak cases, along with previous experience simulating frequency events in the ERCOT system, the DWG did not expect to see system frequencies at or below 59.3 Hz from the simulations for this study. Accordingly, the approach taken with regard to underfrequency firm load shed modeling was to add such relay modeling as needed. Load Modeling Dynamic simulation applications model the sensitivity of loads to changes in voltages via the “ZIP” model, shown below:  2   2   V   V    V   V   P  P  p    p    p  Q  Q q    q    q 0  1V  2 V  3  0 1V  2V  3    0   0     0   0   Where,

P0 = Initial real power Q0 = Initial reactive power V0 = Initial bus voltage p1 = Fraction of P0 for which resistance is constant q1 = Fraction of Q0 for which reactance is constant p2 = Fraction of P0 for which current is constant q2 = Fraction of Q0 for which current is constant p3 = Fraction of P0 that is constant q3 = Fraction of Q0 that is constant P = Real power at a given time step Q = Reactive power at the time step V = Bus voltage at the time step

The factors used in the ZIP model (p1, p2, p3, q1, q2, and q3) for ERCOT loads in this study are based on a 1980’s study conducted by the University of Texas at Arlington (UTA). The ERCOT ZIP model components are shown in table 3. Table 3 - ZIP Model Components Used in ERCOT REACTIVE POWER REAL POWER COMPONENTS COMPONENTS COMPANY p1 p2 q1 q3

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

10 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Oncor 56 44 50 50 Centerpoint 50 50 50 50 CPS 80 20 50 50 AEP (WTU) 41 59 50 50 LCRA 50 50 50 50 AE 50 50 50 50 AEP (CPL) 50 50 50 50 BEC 79 21 50 50 TMPA 79 21 50 50 TNMP2 50 50 50 50 STEC 79 21 50 50

The sensitivity of loads to changes in frequency were modeled via the Siemens PTI library model “LDFRAL”, as shown below: 0, ‘LDFRAL’,’*’,m,n,r,s The model “LDFRAL” modifies the “ZIP” model for all buses in the network model in the following way:

2 r m   V   V        P  P  p    p     p    0  1 V  2 V    3       0   0  0   0  

2 s n   V   V        Q  Q q    q     q    0  1V  2 V    3      0   0  0   0   Where, m = Constant real power load exponent n = Constant reactive power load exponent r = Constant real load current exponent s = Constant imaginary load current exponent

ω0 = Initial bus frequency ω = Bus frequency at a given time step

As in all dynamic studies performed for the ERCOT system in the past, the specific “LDFRAL” model used in this study is as follows: 0, ‘LDFRAL’,’*’,0,2.0,0,0 Wind Generation Modeling The focal point of this study is the system frequency response within the first 20-30 seconds following loss of generation. Since wind plant generation is non- synchronous, the dynamic characteristics of wind turbines have no significant impact

2 The TNMP loads were not part of the UTA research project. Typical characteristics for the TNMP loads were used in this study. This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

11 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

on system frequency response following loss of generation elsewhere as long as most, or all of the wind generation remains online and the wind generation electrical output remains approximately constant over the 20-30 second study interval. The only considerable effect wind plant generation can have on system frequency response in this study occurs when the electrical output of wind turbines changes abruptly over the study interval. For the purposes of this study, abrupt changes in wind plant electrical output arise when wind turbines trip offline due to either frequency or voltage deviations at the terminals of the wind turbine generators beyond thresholds corresponding to the wind turbine generator over/under frequency and over/under voltage relay settings. Therefore, the DWG’s approach to wind modeling in this study was to monitor voltage and frequency at all wind plant buses in the simulations and include loss of wind turbine electrical output in the simulations only as needed. Study Approach Simulation Plan The spinning reserve and LaaRs components of 2800 MW RRS are varied in this study to identify thresholds where reliability concerns emerge. There are, of course, numerous other related variables that must be taken into account. The sensitivities between many of the other variables and power system electrical quantities such as frequency and voltage were not intuitive and, therefore, needed to be simulated. The power system variables analyzed in this study are summarized in Table 4.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

12 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Table 4 – Power System Variables Analyzed VARIABLE COMMENTS Varied between 40% and 75% of 2800 MW RRS. Amount of LaaRs Sum of LaaRs and spinning reserve held at 2800 MW for all simulations. - Governor response allowed for only participating units in most of the simulations. In those cases, varied spinning reserve between 25% and 60% of 2800 MW Amount of spinning reserve/Generator RRS. Sum of LaaRs and spinning reserve held at 2800 governor response. MW for these simulations. - Sensitivity runs were conducted with governor response from all on-line generating units. Amount of generation tripped Varied between 500 MW and 2500 MW. Location of generation tripped Varied between Comanche Peak and South Texas Project. (1) Current locations (approximately 80% along the Gulf Coast, remaining spread approximately even throughout remainder of Texas). (2) Distributed uniformly across the following areas: Location of LaaRs a. Entire ERCOT system. b. West Texas CSC zone. c. North Texas CSC zone. d. Houston CSC zone. e. South Texas CSC zone. 59.7 Hz Tier (1) For current LaaR locations, used current LaaR relay trip settings. LaaR relay trip settings (2) For all other LaaR locations, varied between 59.7 and 59.78 Hz. 59.8 Hz Tier Varied between 59.8 and 59.88 Hz. LaaR relay trip delays Varied between 0 cycles and 20 cycles.

The simulations performed in this study were arranged with the goal of producing results that would be applicable to any operating scenario in the ERCOT system. Accordingly, the simulations were designed to identify the worst case conditions for tripping LaaRs in the ERCOT system and the subsequent limits on the amount of LaaRs that can be reliably deployed. The simulations performed in this study are summarized in Table 5.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

13 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Table 5 – Summary of Simulations GENERATION R

LaaR CONFIGURATION S N E N

TRIPPED I M S E S U L L C S T P A o T E U M S O O B O A E R f W I R M L E N

B ( ( R C C 2 R S T I P A c c V ( O P O T N E T 8 H R E A A E y E I y E T T E

U 0 I R I R c c D z A I R M T N T I

I 0 P S O ) ( N l l K N E I I K e A e M G O

E AMOUNT O O M T N G s s L E T F W N N

) ) S A W ( E R % ) Y ) Note 1 Note 2 Note 3 Note 4 1400- 1 SP 50-75 CUR CUR 15 5 2500 STP 6 700 1400- 2 SP 50-75 CUR CUR 15 5 2300 CP 6 700 3 SOP 65-75 CUR CUR 15 5 980-700 900-1150 CP 18 4 SOP 65 EE 59.70 15 5-15 980 900-1150 CP 18 1400- 5 SOP 50-65 EE 59.70-59.78 15 5 500-1000 CP 120 980 1400- 6 SOP 50-65 CUR CUR 15 5 500-1000 CP 24 980 1400- 7 SOP 50-60 WTX 59.70-59.74 15 5 500-1000 CP 54 1120 1400- 8 SOP 50-60 NTX 59.70-59.74 15 5 500-1000 CP 54 1120 1400- 9 SOP 50-60 HOU 59.70-59.74 15 5 500-1000 CP 54 1120 1400- 10 SOP 50-60 STX 59.70-59.74 15 5 500-1000 CP 54 1120 CUR & CUR & 1400- 11 SOP 50-60 15 5 500-1000 CP 54 WTX 59.70-59.74 1120 CUR & CUR & 1400- 12 SOP 50-60 15 5 500-1000 CP 54 NTX 59.70-59.74 1120 CUR & CUR & 1400- 13 SOP 50-60 15 5 500-1000 CP 54 HOU 59.70-59.74 1120 CUR & CUR & 1400- 14 SOP 50-60 15 5 500-1000 CP 54 STX 59.70-59.74 1120 1400- 15 SOP 50-60 WTX 59.70-59.74 15 5 500-1000 STP 54 1120 1400- 16 SOP 50-60 NTX 59.70-59.74 15 5 500-1000 STP 54 1120 1400- 17 SOP 50-60 HOU 59.70-59.74 15 5 500-1000 STP 54 1120 1400- 18 SOP 50-60 STX 59.70-59.74 15 5 500-1000 STP 54 1120 10011- 19 SOP 50-60 WTX 59.70-59.74 15 5 500-1000 STP 54 10466 10011- 20 SOP 50-60 NTX 59.70-59.74 15 5 500-1000 STP 54 10466 10011- 21 SOP 50-60 HOU 59.70-59.74 15 5 500-1000 STP 54 10466 10011- 22 SOP 50-60 STX 59.70-59.74 15 5 500-1000 STP 54 10466 1400- 23 SOP 50-60 WTX 59.70-59.74 0-10 5 500-1000 STP 162 1120 1400- 24 SOP 50-60 NTX 59.70-59.74 0-10 5 500-1000 STP 162 1120 1400- 25 SOP 50-60 HOU 59.70-59.74 0-10 5 500-1000 STP 162 1120

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

14 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

GENERATION R

LaaR CONFIGURATION S N E N

TRIPPED I M S E S U L L C S T P A o T E U M S O O B O A E R f W I R M L E N

B ( ( R C C 2 R S T I P A c c V ( O P O T N E T 8 H R E A A E y E I y E T T E

U 0 I R I R c c D z A I R M T N T I

I 0 P S O ) ( N l l K N E I I K e A e M G O

E AMOUNT O O M T N G s s L E T F W N N

) ) S A W ( E R % ) Y ) Note 1 Note 2 Note 3 Note 4 1400- 26 SOP 50-60 STX 59.70-59.74 0-10 5 500-1000 STP 162 1120 1400- 27 SOP 50-60 WTX 59.70-59.74 20 5 500-1000 STP 54 1120 1400- 28 SOP 50-60 NTX 59.70-59.74 20 5 500-1000 STP 54 1120 1400- 29 SOP 50-60 HOU 59.70-59.74 20 5 500-1000 STP 54 1120 1400- 30 SOP 50-60 STX 59.70-59.74 20 5 500-1000 STP 54 1120 55 59.74 1120- 31 SOP EE 15 5 500-1250 STP 180 60-75 59.80-59.88 700 50 59.72 1260- 32 SOP WTX 15 5 500-1250 STP 225 55-75 59.80-59.88 700 55 59.72 1120- 33 SOP NTX 15 5 500-1250 STP 180 60-75 59.80-59.88 700 50 59.74 1260- 34 SOP HOU 15 5 500-1250 STP 225 55-75 59.80-59.88 700 55 59.72 1120- 35 SOP STX 15 5 500-1250 STP 180 60-75 59.80-59.88 700 1400- 36 SOP 50-75 CUR MCUR 15 5 500-1250 STP 63 700 1400- 37 SOP 50-75 CUR 59.70-59.78 15 5 500-1250 STP 270 700 Note 1: SP = Summer Peak SOP = Spring Off-Peak Note 2: CUR = Located at current contracted locations EE = Uniformly over entire ERCOT system WTX = Uniformly over West Texas CSC zone NTX = Uniformly over North Texas CSC zone HOU = Uniformly over Houston CSC zone STX = Uniformly over South Texas CSC zone CUR & WTX = 50% of 2300 MW at current contracted locations with remaining uniformly over West Texas CSC zone CUR & NTX = 50% of 2300 MW at current contracted locations with remaining uniformly over North Texas CSC zone CUR & HOU = 50% of 2300 MW at current contracted locations with remaining uniformly over Houston CSC zone CUR & STX = 50% of 2300 MW at current contracted locations with remaining uniformly over South Texas CSC zone Note 3: CUR = Tripped at current relay trip set-points. 9.1% of LaaRs tripped at 59.8 Hz, 0.9% at 59.72 Hz, 1.4% between 59.71 and 59.72 Hz, 10.7% at 59.71 Hz, 25.4% between 59.70 and 59.71 Hz, 52.5% at 59.7 Hz. MCUR = Modified version of current relay trip set-points. 0.9% LaaRs tripped at 59.72 Hz, 1.4% between 59.71 and 59.72 Hz, 10.7% at 59.71 Hz, 25.4% between 59.70 and 59.71 Hz, 61.6% at 59.7 Hz. Note 4: STP = South Texas Project CP = Comanche Peak

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

15 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Other Simulation Details The following output channels were included for all simulations: 1. Frequency and voltage at the Jewett 345 kV bus. 2. Bus voltage angle statistic channels. 3. Rotor angles for all machines. 4. Complex power for all machines. 5. Terminal voltage for all machines. 6. Field voltage for all machines. 7. Various power statistics for all areas and zones. 8. Frequency and voltage at all wind plants. A 30 second simulation period was used for all simulations. Study Criteria There were several criteria used in the interpretation of the study results. 1. System stability – For a scenario to be considered acceptable, there must be no abnormal system conditions such as machines out-of-step. 2. Underfrequency limit – As described in the above section on low set underfrequency load shed relay modeling, initial runs were conducted without such relays modeled. The underfrequency limit for all of these initial runs was 59.3 Hz. This limit was used only as a threshold to model underfrequency firm load shed as needed. 3. Overfrequency limit (frequency overshoot) – The maximum operating frequency for a power system is a function of numerous factors, the most notable of which are related to how the major components of generating plants are affected by overfrequency. The overfrequency limit of any power system, in terms of both frequency and duration of overfrequency, is most likely to be established by the overfrequency relaying protecting certain power plant components. Because the capabilities of generating plant equipment varies widely, there are no national standards that establish limits related to overfrequency ride-through for relaying at generating plants. However, some reliability organizations do establish such limits. For instance, the Western Electricity Coordinating Council requires the following with regard to generator overfrequency relaying3: - Continuous operation for frequencies between 60 Hz and 60.6 Hz - Minimum trip time of 3 minutes for frequencies > 60.6 Hz. - Minimum trip time of 30 seconds for frequencies > 61.6 Hz. - Instantaneous trip allowed for frequencies above 61.7 Hz.

3 “WECC Coordinated Off-Nominal Frequency Load Shedding and Restoration Requirements”, July 2005. This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

16 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

In the absence of such limits for the ERCOT system, the DWG selected 60.4 Hz as the overfrequency limit for this study. The 60.4 Hz overfrequency criterion is a conservative limit used in the responsive reserve study conducted by the DWG in 2002. Increasing the overfrequency limit above 60.4 Hz should not be done without coordination with plant operators and a detailed study of plant equipment in the ERCOT system. 4. Wind plant voltage and frequency – Voltage and frequency were monitored at all wind plant buses in order to properly model any loss of wind plant generation due to under/over voltage and/or under/over frequency conditions. Wind generator voltage relaying is typically set for continuous operation between 90% and 110% of nominal voltage. For loss of wind generation to occur due to a voltage issue within the 30 second simulation interval used in this study, wind generator terminal voltage would typically need to drop to about 30% or rise above about 110% of nominal voltage. Wind generator frequency relaying is typically set such that the unit can operate continuously between about 57 Hz and 62 Hz. For frequencies outside of that band, the frequency relaying typically trips instantaneously. Subsequently, the following voltage and frequency criteria were used for all wind plant buses to determine the need for modeling loss of wind plant generation: - Undervoltage limit of 30% of nominal. - Overvoltage limit of 110% of nominal - Underfrequency limit of 57 Hz. - Overfrequency limit of 62 Hz. STUDY RESULTS Summer Peak Conditions During peak loading conditions, system inertia is at or near peak levels. Therefore, frequency overshoot following tripping of LaaRs during the summer peak was either not expected to be excessive or not expected to occur. The series 1 and 2 simulations sufficiently confirm these expectations. No frequency overshoot occurs in any of the summer peak scenarios. The minimum frequency drop, about 59.63 Hz, occurred for 2500 MW loss of the STP plant with the minimum amount of LaaRs simulated (50% of 2800 MW, or 1400 MW). Spring Off-Peak Conditions Since system inertia is the lowest during off-peak loading, the Spring off-peak scenarios are the critical cases for frequency overshoot. The following paragraphs summarize the series 3-35 Spring off-peak simulations and results. Series 3 In the first series of simulations for the Spring off-peak case, existing LaaR relay locations and corresponding trip settings are modeled to screen out the larger percentages of LaaRs that would result in excessive frequency overshoot. Overshoot violations occur with LaaRs percentages greater than 65% when the amount of generation lost is in the range of 1000 – 1050 MW.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

17 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Series 4 In the series 4 simulations, LaaRs are held at 65% while the LaaR breaker operate times were varied. Furthermore, two additional aspects of the LaaR data were varied moving from the series 3 to series 4 simulations. The first was the LaaR locations. In the series 4 simulations, all of the LaaRs are spread uniformly across the entire ERCOT system. The second was the LaaR trip settings. All LaaRs were set to trip at 59.7 Hz in series 4. The series 4 plots indicate the variance of breaker operate times have no bearing on frequency response. Although there were no frequency overshoot violations in the series 4 runs, the series 4 plots reveal two other important details. 1. There is an upward trend in frequency overshoot as the amount of generation tripped is varied towards the minimum simulated 900 MW. 2. The series 3 simulations did not show an upward trend in frequency overshoot as the amount of generation tripped is decreased. This is because existing LaaR relay data is used in the series 3 runs. The trip levels in the existing LaaR data is such that approximately 9% of the total LaaRs trips at 59.8 Hz, 1% LaaRs trips at 59.72 Hz, and the remaining LaaRs trips at frequencies between 59.7 and 59.72 Hz. Having the LaaRs spread out over a range of frequencies, particularly having 9% of the LaaRs tripping at 59.8 Hz and all of the remaining LaaRs tripping at 59.72 or less, effectively blunts frequency overshoot. This critical detail supports the notion that existing LaaR data should not be used to determine LaaRs limits applicable in a general sense. While the existing LaaR data may be realistic for the present, the existing LaaR data may not always be applicable. Based on these two observations, most of the successive runs test system frequency response from tripping all LaaRs at the same frequency following loss of generation less than 900 MW. Series 5 Results from the series 5 runs support conclusions made from comparison of the series 3 and 4 results. In this series LaaRs percentages above 55%, as well as 55% LaaRs tripped above 59.74 Hz, result in frequency overshoot violations. Series 6 The purpose of the series 6 runs was to have another means of verifying the conclusion made from the series 3 and series 4 runs that use of the existing LaaR data produces optimistic results. The only modeling detail changed going from the series 5 runs to the series 6 runs was the use of existing LaaR data. Results from the series 6 runs confirm the aforementioned conclusion. There were no frequency overshoot violations in the series 6 runs. Series 7-10 In all of the previous simulations, the geographical nature of the LaaRs was such that either the LaaRs were distributed uniformly throughout the entire ERCOT system or the existing LaaR geographical distribution was modeled. The purpose of the series 7-10 simulations was to determine the relationship, if any, between system frequency response and location of LaaRs and, assuming a measurable relationship exists, to This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

18 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

determine whether or not the series 5 runs were the worst case scenario in terms of LaaR geographical locations. All LaaRs were distributed uniformly in the West Texas, North Texas, Houston, and South Texas Commercially Significant Constraint (CSC) zones in the series 7, 8, 9, and 10 runs, respectively. The following LaaRs percentages resulted in frequency overshoot violations in the series 7-10 runs: o 55% or greater when located in the West Texas CSC zone. o Greater than 55% when located in the North Texas CSC zone. o 55% or greater when located in the Houston CSC zone. o Greater than 55% when located in the South Texas CSC zone. Comparing the series 7-10 results to the series 5 results, there is a small sensitivity of frequency overshoot to the geographical locations of LaaRs. Having the LaaRs distributed uniformly throughout the ERCOT system presents a slightly more optimistic scenario than confining the LaaRs to the various CSC zones. Series 11-14 The purpose for the series 11-14 runs was to examine the present-day situation of adding additional LaaRs to the existing LaaR pool. Existing LaaRs totaling 50% of 2300 MW, or 1150 MW, are modeled in all of the runs. Additional LaaRs needed for modeling 55%-60% LaaRs are lumped in the West Texas, North Texas, Houston, and South Texas CSC zones in the series 11, 12, 13, and 14 runs, respectively. The series 11-14 runs further verify the conclusion that the frequency trip settings in the existing LaaR data yield optimistic results that should not be used to determine LaaR limits in a general sense. Again, about 9% of the existing LaaRs is tripped at 59.8 Hz and the remaining existing LaaR is tripped at or below 59.72 Hz. Subsequently, there were no frequency overshoot violations in the series 11-14 results. Series 15-18 In the series 3-14 simulations, loss of generation at the Comanche Peak plant was simulated. Although the inertia of the two Comanche Peak generating units are relatively high, the inertia of these two units are not the highest within the ERCOT system. The two generating units at South Texas Project (STP) have the highest inertia within the ERCOT system. The purpose of the series 15-18 runs was to examine how the results of the series 7-10 runs would be affected by simulating loss of the higher inertia generation at STP. The series 15-18 results show tripping STP generation presents more restrictive LaaR limits than identified in the previous simulations where Comanche Peak generation was tripped. The following LaaRs percentages and associated frequency trip levels resulted in frequency overshoot violations in the series 15-18 runs: o Greater than 50% and 50% tripped at greater than 59.72 Hz when located in the West Texas CSC zone. o Greater than 55% and 55% tripped at greater than 59.72 Hz when located in the North Texas CSC zone.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

19 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

o Greater than 50% and 50% tripped at greater than 59.74++4 Hz when located in the Houston CSC zone. o Greater than 55% and 55% tripped at greater than 59.72 Hz when located in the South Texas CSC zone. Series 19-22 In the series 3-18 runs, the only generating units responding following loss of generation were the units participating in the responsive reserve service. The series 19-22 runs test the effect of response from all available units in the ERCOT system. The series 19- 22 runs are effectively the series 15-18 runs with the original governor models included for all units not participating in the responsive reserve service. As the series 19-22 results indicate, having additional governor response outside from that of the spinning reserve units lessens frequency overshoot to a great extent. The primary reason for this is the additional generator governor action reduces the rate of frequency decline following loss of generation, reducing the amount of LaaRs that trip. There were no frequency overshoot violations in the series 19-22 runs. Series 23-30 The LaaR relay delay time was simulated at 15 cycles in all of the previous simulations. The purpose of the series 23-30 runs was to test the effect of varying LaaR trip delays in the series 15-18 runs. In these runs, the LaaR trip delays were varied between 0 cycles (instantaneous trip) and 20 cycles. The series 23-30 results show, for the most part, changes in LaaR relay trip delays have no effect on system frequency performance. The exception to this is a few of the scenarios where all of the LaaRs are located in the South Texas CSC zone. The following LaaRs percentages resulted in frequency overshoot violations in the series 23-30 runs that were not already identified in the series 15-18 runs: o Greater than 50% when located in the South Texas CSC zone. Series 31-35 The series 31-35 runs test the deployment of LaaRs in two tiers. The first tier is the same “59.7 Hz” tier studied in the series 1-30 runs. The second tier is a “59.8 Hz” tier. The assumption made in these runs is the amount of LaaRs at the first tier and the corresponding first tier trip levels are held at the applicable threshold identified from the series 5 simulations (setup for series 31 runs) and series 15-18 simulations (setup for series 32-35 runs). There are two key observations from the series 31-35 results: 1. Tiers of LaaRs above the first LaaRs tier cannot be reliably deployed when the first tier LaaRs are operated at threshold levels. In the set-up for the series 32, 33, and 35 runs, the first tier LaaRs are operated at threshold levels. That is, the amount of first tier LaaRs is at a critical level and all first tier LaaRs tripping is done at the same critical frequency. All levels of second tier LaaR tripping resulted in excessive frequency overshoot in the series 32, 33, and 35 runs.

4 The highest LaaR trip level simulated in these runs was 59.74 Hz. The actual threshold is somewhat higher than 59.74 Hz. This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

20 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

2. Tiers of LaaRs above the first LaaRs tier can be reliably deployed when the first tier LaaRs are operated within critical levels. The maximum amount and maximum tripping level of higher tier LaaRs is a function of how far within critical levels the first tier LaaRs are limited. In the set-up for the series 31 runs, the first tier LaaRs are modeled to operate somewhat below actual threshold levels. The amount of first tier LaaRs and tripping frequency used in the series 31 runs are established from thresholds derived from the series 5 runs (55% LaaRs, all tripped at 59.74 Hz). It can be concluded from comparison of the various series 5 plots that the LaaRs threshold tripping frequency for the series 5 runs is actually somewhere between 59.74 and 59.76 Hz. Consequently, the series 31 results indicate second tier LaaRs can be deployed in the following ways: - Up to 5% tripped between 59.82 Hz and 59.8 Hz. - Up to 15% tripped at no more than 59.8 Hz. The same situation exists for the series 34 runs. The first tier LaaRs for the series 34 runs are also modeled to operate below actual threshold levels. The first tier LaaR thresholds used in the series 34 runs are based on thresholds defined in the series 17 runs (50% LaaRs, all tripped at 59.74 Hz). The actual threshold is slightly above 59.74 Hz. Consequently, the series 34 results indicate up to 5% second tier LaaRs, tripped at no more than 59.8 Hz, can be reliably deployed. Series 36 The series 3 runs provide meaningful results for the consideration of the amount of LaaRs that can be reliably deployed assuming the LaaR frequency set-points remain very similar to that of current LaaRs wherein the LaaRs trip over a range of frequencies between 59.7 Hz and 59.78 Hz. Recall the following points from the above discussion on the series 3 and series 4 simulation results: 1. The series 3 runs suggest up to 65% of 2800 MW LaaRs could be reliably deployed assuming the LaaRs maintained a trip frequency and geographic location profile very similar to that of the current LaaR pool. 2. From the comparison between the series 3 and series 4 results, the current LaaR relay frequency set-points produce results that are optimistic relative to scenarios where there exists a tighter band of trip frequencies among the LaaR relays. 3. About 9% of the LaaRs modeled in the series 3 simulations trip at 59.8 Hz. The purpose of the series 36 runs was to quantify how much the 65% max LaaRs number would decrease when the relay trip set-point of the 9% LaaRs, set to trip at 59.8 Hz in the series 3 runs, was changed to a lower frequency. Specifically, the relay trip set-point for the 9% LaaRs was changed from 59.8 Hz to 59.70 Hz in the series 36 runs. Plots from the series 36 runs indicate that frequency overshoot violations will occur with LaaRs percentages greater than 60%.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

21 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

Series 37 The series 37 runs build on the series 36 runs. The purpose of the series 37 runs was to identify the maximum allowable amount of LaaRs, assuming the geographical locations of the LaaRs are similar to that of the current LaaR pool, when all of the LaaRs trip at the same frequency. The series 37 runs essentially identify the worst case scenario for tripping LaaRs when the geographical locations of the LaaRs are similar to that of the current LaaR pool. Results from the series 37 runs demonstrate that up to 55% of 2800 MW LaaRs, geographically distributed similar to the current LaaR pool, can be reliably deployed with all of the LaaRs set to trip at no more than 59.72 Hz.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

22 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008

CONCLUSIONS Because system inertia is at the lowest levels during light loading, conditions are optimal for system frequency overshoot. Consequently, the critical operating scenarios for reliable deployment of LaaRs are during periods of light loading. The most significant variables to consider in determining the maximum amount of LaaRs that can be deployed via underfrequency relaying are the frequency trip settings of the LaaR relays. The worst case scenarios for frequency overshoot occur when all of the LaaRs trip at the same frequency set- point. Intermediately, frequency overshoot can be depressed to varying degrees when the frequency set-points of the LaaR relays are spread out over a range of frequencies. Other variables having a more limited relationship with the maximum amount of LaaRs deployable by underfrequency relaying are the geographical locations of the LaaRs. Table 6 summarizes the various limits on the amount of LaaRs deployable at the 59.7 Hz tier and the conditions associated with each limit. Table 6 – Summary of LaaRs Limits at the 59.7 Hz Tier RELAY TRIP SETTING LaaRs LIMIT (% of 2800 LOCATION OF LaaRs LIMIT BASIS (Hz)† MW) Current trip settings: 9.1% @59.8 Hz 0.9% @59.72 Hz 1.4% between 59.71 and Current locations 59.72 Hz 65 10.7% @59.71 Hz 25.4% between 59.70 and 59.71 Hz 52.5% @59.70 Hz Modified version of the current trip settings: 0.9% @59.72 Hz 1.4% between 59.71 and Overshoot greater than Current locations 59.72 Hz 60 or equal to 60.4 Hz. 10.7% @59.71 Hz 25.4% between 59.70 and 59.71 Hz 61.6% @59.70 Hz Current locations 59.72 55 Uniformly distributed 59.74+ 55 throughout ERCOT Lumped in NTX CSC zone 59.72 55 Lumped in STX CSC zone 59.74 50 Lumped in WTX CSC zone 59.72 50 Lumped in HOU CSC zone 59.74+ 50 †The relay trip setting data in row 1 of this table is representative of the current LaaRs in ERCOT. The relay trip setting numbers in rows 3 through 8 are maximum LaaR trip frequencies assuming all of the LaaRs specified in the “LaaRs LIMIT” column are tripped at the same frequency. Results from simulations of LaaRs deployed at the 59.8 Hz tier indicate such LaaRs cannot be reliably deployed with the 59.7 Hz tier LaaRs operated at the critical levels identified in table 6. The maximum amount of LaaRs and corresponding maximum relay trip settings that can be reliably deployed at the 59.8 Hz tier is a function of where limits are established for the amount of LaaRs and corresponding relay trip settings at the 59.7 Hz tier. Therefore, the DWG will This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

23 DRAFT - 2008 ERCOT Loads Acting as a Resource Capability Study December, 2008 need guidance from ROS on where the 59.7 Hz tier limits will be set in order to define LaaR limits for the 59.8 Hz tier. Section 6.10.3.2 of the ERCOT Protocols allow a LaaR to be deployed at up to 150% of the amount requested by ERCOT at the time of testing the LaaR. This potential variance between contracted and actual LaaR amount should be considered when establishing LaaR limits.

This document contains proprietary information and shall not be reproduced in whole or in part without prior written permission of ERCOT

24